Article View: comp.sys.mac
Article #23367Re: Stats on 68881
From: Steve Hix
Date: Wed, 06 Jan 1988 20:50
Date: Wed, 06 Jan 1988 20:50
26 lines
1218 bytes
1218 bytes
In article <8368@steinmetz.steinmetz.UUCP>, (Jeff A. Hallett) writes: > > Does anyone know how fast the 68881 can do a multiply-add? Well, it depends... What clock rate is your 68881 running at? What is the format of the memory source or destination operand? What addressing mode are you using? (If I'm reading one of the tables in the "MC68881 Floating- Point Coprocessor User's Manual" rightly, the FMUL instruction can take anywhere from 71 to 908 cycles to execute. This don't allow for instruction/data fetch and write time... It might be easier to just find a Mac with the chip in it and write your own benchmark to exercise the '881 and find a typical time for the application sort of thing you're doing. (But we all know about bench- marks, don't we? :} ) Or you could get the Motorola '881 manual (MC68881UM/AD) and wallow around in Chapter 6 for a while. seh
Technical Details
Message-ID:
<38058@sun.uucp>Path:
arc1-ca.pugleaf.net!utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!concertina!fiddlerReferences:
<8368@steinmetz.steinmetz.UUCP>Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!concertina!fiddler From: fiddler%concertina@Sun.COM (Steve Hix) Newsgroups: comp.sys.mac Subject: Re: Stats on 68881 Message-ID: <38058@sun.uucp> Date: 6 Jan 88 20:50:42 GMT References: <8368@steinmetz.steinmetz.UUCP> Sender: news@sun.uucp Lines: 26